Expansion of human bone marrow-derived mesenchymal stromal cells with enhanced immunomodulatory properties

Background Mesenchymal stromal cells (MSCs) have broad potential as a cell therapy including for the treatment of drug-resistant inflammatory conditions with abnormal T cell proliferation such as graft-versus-host disease (GVHD). Clinical success, however, has been complicated by the heterogeneity of culture-expanded MSCs as well as donor variability. Here, we devise culture conditions that promote expansion of MSCs with enhanced immunomodulatory functions both in vitro and in animal models of GVHD. Methods Human bone marrow-derived MSCs were expanded at high-confluency (MSCHC) and low-confluency state (MSCLC). Their immunomodulatory properties were evaluated with in vitro co-culture assays based on suppression of activated T cell proliferation and secretion of pro-inflammatory cytokines from activated T cells. Metabolic state of these cells was determined, while RNA sequencing was performed to explore transcriptome of these MSCs. Ex vivo expanded MSCHC or MSCLC was injected into human peripheral blood mononuclear cells (PBMC)-induced GVHD mouse model to determine their in vivo therapeutic efficacy based on clinical grade scoring, human CD45+ blood count and histopathological examination. Results As compared to MSCLC, MSCHC significantly reduced both the proliferation of anti-CD3/CD28-activated T cells and secretion of pro-inflammatory cytokines upon MSCHC co-culture across several donors even in the absence of cytokine priming. Mechanistically, metabolic analysis of MSCHC prior to co-culture with activated T cells showed increased glycolytic metabolism and lactate secretion compared to MSCLC, consistent with their ability to inhibit T cell proliferation. Transcriptome analysis further revealed differential expression of immunomodulatory genes including TRIM29, BPIFB4, MMP3 and SPP1 in MSCHC as well as enriched pathways including cytokine–cytokine receptor interactions, cell adhesion and PI3K-AKT signalling. Lastly, we demonstrate in a human PBMC-induced GVHD mouse model that delivery of MSCHC showed greater suppression of inflammation and improved outcomes compared to MSCLC and saline controls. Conclusion Our study provides evidence that ex vivo expansion of MSCs at high confluency alters the metabolic and transcriptomic states of these cells. Importantly, this approach maximizes the production of MSCs with enhanced immunomodulatory functions without priming, thus providing a non-invasive and generalizable strategy for improving the use of MSCs for the treatment of inflammatory diseases. Supplementary Information The online version contains supplementary material available at 10.1186/s13287-023-03481-7.

immunomodulatory functions without priming, thus providing a non-invasive and generalizable strategy for improving the use of MSCs for the treatment of inflammatory diseases.

Background
Mesenchymal stromal cells (MSCs) have been widely investigated as a cell therapy product due to their low immunogenicity, robust ability to engraft and differentiation potential into osteoblasts, adipocytes and chondrocytes [1].MSCs can be isolated from bone marrow [2], adipose tissue [3], placenta [4] and umbilical cord [5] making them a readily available cell type for clinical indications; however, their utility has been limited by donor variability and their unpredictable phenotypic heterogeneity during ex vivo expansion.For example, the yield of MSCs isolated from primary tissue is often low and requires expansion to reach sufficient numbers for therapy.A lack of standardized cell culture conditions including cell seeding density could introduce further heterogeneity in culture-expanded MSCs leading to differences in morphology, gene expression patterns and differentiation potency [6][7][8].MSC enrichment is rarely performed after culture expansion before administration to a patient, which can also impact their therapeutic potency.
Growing evidence suggests that MSCs are a promising therapy for treating inflammatory conditions and diseases due to their immunomodulatory effects [9][10][11][12][13][14].Studies show that MSCs can suppress inflammation upon migrating to sites of damage when infused in vivo [15].Although the mechanisms of action are still unclear, MSCs can inhibit proliferation of activated T cells by promoting cell cycle arrest [9,11,[16][17][18].Consequently, immunomodulation of T cells by MSCs results in reduced secretion of the pro-inflammatory cytokines IFN-γ and IL-17A.Despite their potential, a lack of data connecting biological activity to therapeutic outcomes has limited the use of MSC for inflammatory diseases [19,20] including graft-versus-host disease (GVHD) [21,22] and rheumatoid arthritis [23,24].Thus, ex vivo expansion strategies that can enrich for MSCs with defined immunomodulatory functions have become a major focus to improve their use as cell therapy products.
Currently, priming MSCs with pro-inflammatory cytokines including TNF-α [25] and IFN-ɣ [26,27] are used to improve their ability to suppress T cells and reduce inflammation.Cytokine priming strategies, however, have several disadvantages when considering downstream clinical applications.For example, priming can lead to increased immunogenicity of MSCs [28] and possible host rejection, reducing safety and efficacy.Priming can also introduce further heterogeneity during expansion given that different strategies employ various combinations and concentrations of cytokines [25,29].Moreover, the use of recombinant cytokines is costly and requires timely characterization due to batch differences and specific activities during production [30].Finally, current methods that use unselected peripheral blood mononuclear cells (PBMCs) to study MSC-mediated immunomodulation could make it difficult to distinguish the direct effects of MSCs on T cells due to the presence of other immune cell types [18,[31][32][33].
Here, we show that culturing bone marrow-derived MSCs cultured at high confluency (MSC HC ) without cytokine priming robustly enhanced the suppression of anti-CD3/CD28-activated T cell proliferation and the secretion of pro-inflammatory cytokines across multiple donors.MSC HC conditions resulted in metabolic reprogramming towards glycolysis and to increased lactate production, properties known to limit T cell proliferation.Consistent with our observations, transcriptome analysis in MSC HC revealed differential expression of genes and pathways associated with immunomodulation compared to MSC LC .Specifically, we identified enriched pathways related to immunomodulation such as cytokine-cytokine receptor interactions, cell adhesion and PI3K-AKT signalling in MSC HC donors exhibiting the most robust immunosuppression in co-cultures.GVHD is a known challenge to allogenic hematopoietic stem cell transplant, and it primarily involves donor T cells in the graft attacking host tissues, such as skin, gastrointestinal tract and liver [34].The standard care is often steroids, but about half of the patients develop steroid resistance [35].Treatment of human PBMCs-induced GVHD mouse model with MSC HC led to a marked reduction in GVHD symptoms as compared to MSC LC -treated mice.Specifically, we show that a single MSC HC dose at the onset of symptoms resulted in a lower disease severity based on clinical grade scoring and reduced lymphocyte infiltration in major organs.Thus, our approach provides a generalizable and inexpensive strategy that can be readily adapted to improve the safety and efficacy of basal MSCs for the treatment of GVHD and potentially more broadly for inflammatory disorders.

BM-MSC culture
Human adult bone marrow-derived MSCs were obtained from healthy donors (n = 6) and commercially purchased from Lonza Pte.Ltd (Lonza PT-2501).Donor information of these 6 donors is given in Table 1.
These cells were cultured in MSC expansion media (low-glucose Dulbecco's modified Eagle medium (DMEM) (Gibco) supplemented with 10% foetal bovine serum (FBS) (Gibco) and 1% penicillin/streptomycin (Gibco).All cells were thawed from frozen vials and subjected to recovery.They were routinely maintained in a humidified 37˚C incubator with 5% CO 2 , at a seeding density of 200 cells/cm 2 (low confluency) or 3000 cells/cm 2 (high confluency).Culture media was changed every 3 days and were harvested at same time using 0.05% Trypsin/EDTA solution (Life Technologies).The cell count and viability were determined using disposable haemocytometer (INCYTO, KR) and Trypan Blue (Thermo Fisher Scientific, SG).Cell confluency was quantified from a set of phase contrast microscopy images taken at 4× magnification (Olympus SC30) using a ImageJ algorithm (SMART CAMP).Cells harvested at end of passage 4 were used for all downstream experiments in this study.

In vitro multi-lineage differentiation of MSCs
MSCs were seeded at density of 3 × 10 4 cells/well and 6 × 10 4 cells/well for osteogenesis and adipogenesis, respectively, in 24-well plates for overnight attachment.The MSC expansion media was removed the following day and replaced with specific differentiation media (STEMCELL Technologies).Differentiation media was changed every 3 days for 14 days before processing.Osteogenesis induction was confirmed with Alizarin Red S staining (ScienCell) for calcium deposits while adipogenic differentiation was determined with Oil Red O staining (Sigma) for detection of lipid droplets according to manufacturer's protocol.The subsequent dye extraction of Oil Red O and Alizarin Red S was performed for measurement of optical densities (OD) at 405 nm and 510 nm, respectively, using a plate reader (Tecan).Quantification of adipogenesis and osteogenesis was calculated by the measured OD of stain divided by the seeded cell number of each sample, represented as the normalized OD.
For chondrogenesis, MSCs were pelleted at 0.5 × 10 6 cells/pellet in a 15-ml tube by centrifugation at 300g, 5 min.Chondrogenic media (STEMCELL Technologies) was added to this pellet and media was changed every 3 days for 3 weeks before glycosaminoglycan quantification.Hydrogel constructs were fixed in 10% neutral buffered formalin (Sigma-Aldrich) overnight at 4 °C followed by graded ethanol dehydration (70%, 70%, 80%, 80%, 95%, 95% ethanol at 15 min each) and 3 min of incubation with eosin (Sigma-Aldrich) for easy visualization during tissue sectioning.The samples were then washed with 100% ethanol thrice and dehydrated further in 100% ethanol for 20 min twice and HistoChoice ® clearing agent (Sigma-Aldrich) for 20 min twice.The dehydrated samples were incubated in paraffin bath for 1 h before embedding in paraffin.Samples were cut into 5 µm section using microtome (Leica, USA) and collected on silane-coated slides (Thermo Scientific, USA).Tissue sections were de-waxed and rehydrated in xylene and graded ethanol, respectively.Immunohistochemical staining was used to identify Type II Collagen using Ultra Vision detection kit (Thermo Fisher Scientific, SG).The rehydrated sections were first blocked with hydrogen peroxide block for 15 min before pepsin treatment for 20 min at 37 °C.After which, the sections were treated with Ultra V Block for 5 min and subsequently with 1-h primary antibody incubation using Type II Collagen mouse monoclonal antibodies (Clone 6B3 at 1:500 dilution, Sigma-Aldrich).It was followed by secondary antibody incubation with biotinylated goat anti-mouse secondary antibody for 30 min.Streptavidin peroxidase was then added and incubated for 45 min before 1.5 min of incubation with DAB chromogen.Lastly, the sections were counterstained with hematoxylin, dehydrated and mounted for imaging.

In vitro T cell assay
MSCs were seeded at 5 × 10 4 cells/well in 48-well plate overnight before assay.PBMCs from a single donor were commercially purchased (STEMCELL Technologies) with details in Table 2.
Negative immunoselection was performed to isolate CD3 + T cells using the EasySep ™ Human T Cell Isolation Kit (STEMCELL Technologies) according to the manufacturer's instructions.The purity of CD3 + T cells was more than 95%.CD3 + T cells were stained with Celltrace Violet (CTV) dye (Thermo) and stimulated with ImmunoCult ™ Human CD3/CD28 T cell activator (STEMCELL Technologies) before direct co-culture with MSCs for 5 days at 1:3 (MSC: T cell ratio).These activators contain anti-human CD3 monospecific antibody complex and anti-human CD28 monospecific antibody complex.The co-culture media was collected at the end of day 5 of co-culture and spun at 1000 g, 5 min to retrieve the cell pellet (T cells) while the supernatant was kept in − 80 °C until cytokine analysis.Flow cytometric analysis was performed on harvested T cells using fluorochrome-conjugated anti-human antibodies: CD3-APC, CD4-BV605, CD45-PerCP Cy5.5, CD8a-PE-Cy7 (BD Biosciences).Positively stained cell populations were identified by comparing with FMO controls.Acquisition was performed using a CytoFLEX flow cytometer (Beckman Coulter Life Sciences) and analysed using FlowJo V10.Quantification of T cell proliferation was determined using the formula [36] to determine proliferative index (PI): PI = Log[FI nd /MFI all ]/Log [2], with MFI all = median fluorescence intensity of all viable T cells and FI nd = peak fluorescence intensity of the viable non-divided cells.

Enzyme-linked immunosorbent assay (ELISA)
The concentration of cytokines, including IFN-ɣ and IL-17A, was determined by Luminex ® -based multiplex assays (R&D Systems) according to manufacturer's instructions.All samples were analysed in duplicate and repeated for 6 different donors.The data were obtained with a MAGPIX reader (Millipore), and concentrations were derived from measured mean fluorescence intensities using fitted standard curves using 5-parameter logistic regression (SSL5) using the Milliplex Analyst software (Millipore).

Measurement of cell metabolism
MSCs were seeded at 2 × 10 4 cells/well in a Seahorse XF24 V7-PS Microplate (Agilent Technologies) overnight.Cell culture media was collected the next day for measurement of lactate concentration using Cedex Bio Analyzer (Roche).Lactate production rate was calculated by deducting the lactate concentration in base media and normalized to total cell number and time of culture.MSCs (2.5 × 10 4 cells/well) seeded on Seahorse 24-well microplate were used to measure extracellular acidification rate (ECAR) using Seahorse XF24 Extracellular Flux Analyzer (Agilent Technologies) according to manufacturer's protocols.

RNA isolation and sequencing
Total RNA was isolated using RNeasy Mini Kit (Qiagen).RNA libraries were prepared using TruSeq Stranded mRNA Library Prep Kit according to manufacturer's protocol.Transcriptome sequencing was performed using Illumina Novaseq 6000 system.The raw data were analysed using Partek ® Flow ® software (Partek Inc., St. Louis, MO, USA).Reads were mapped to the Genome Reference Consortium Human Reference 38 (hg38) using the STAR aligner.The aligned reads were quantified and annotated based on Ensembl transcripts (release 108).ANOVA was performed as the statistical method for differential expression analysis, and differentially expressed genes (DEGs) were identified by filtering with two criteria for significance: magnitude of fold change (FC) ≥ 2 and false discovery rate step-up ≤ 0.05.Heatmaps were generated using Partek ® Flow ® software with genes mapped according to their respective z-score.Significantly enriched pathways were identified with the Kyoto Encyclopedia of Genes and Genomes Pathways (Homo sapiens) (KEGG) database, with the plot of significantly enriched pathways (p ≤ 0.05) generated by R Studio.

In vivo testing with human PBMC-induced mouse GVHD model
Female immunodeficient mice NOD-SCID IL2rg null (NSG) mice were purchased from the Jackson Laboratory (stock number 005557).All mice were bred and kept under pathogen-free conditions in Biological Resource Centre, Agency for Science, Technology and Research, Singapore (A*STAR) on controlled 12-h light-dark cycle with ad libitum access to normal diet and drinking water.All experiments and procedures were approved by the Institutional Animal Care and Use Committee of the

Immunophenotyping of peripheral blood by flow cytometry
Peripheral blood was collected submandibularly from mice in K3EDTA minicollect ® tubes (Grenier Bio-One) at intermediate time points and endpoint.Red blood cells (RBCs) were lysed using RBC lysis buffer (Life Technologies).Live immune cells from peripheral blood were determined by staining with live/dead fixable blue dead cell stain kit (Life Technologies) for 10 min at room temperature prior to staining with mouse and human Fc receptor blocking reagents (Biolegend) for 10 min at room temperature to prevent non-specific binding.Subsequently, cells were stained with fluorescent-labelled surface markers which include anti-human CD45 (HI30; BD Biosciences) and anti-mouse CD45.1 (A20; Biolegend) for 30 min at 4 °C in the dark.After staining, cells were washed and resuspended in FACS buffer containing PBS, 0.2% bovine serum albumin (GE Healthcare Life Sciences) and 0.05% sodium azide (Merck) for flow cytometry data acquisition.Data were acquired using a LSR Fortessa X20 flow cytometer (BD Biosciences) with FACSDiva software and analysed using FlowJo software (version 10; Tree Star Inc).The absolute count of each immune cell type is determined using CountBright Counting Beads (Thermo Fisher Scientific) following manufacturer's instructions.

Hematoxylin and eosin (H&E) staining
Mice were euthanized using carbon dioxide (CO 2 ) inhalation at endpoint.Lungs, colon, liver, kidneys and skin were removed from euthanized mice and fixed in 10% formalin.Fixed tissues were embedded in paraffin wax, processed to obtain 5 μm sections and subjected to Hematoxylin & Eosin (H&E) (Sigma) staining following established protocols.Histopathological images were acquired using Axio Scan.Z1 slide scanner (Zeiss) and analysed using Zen 2 (blue edition; Zeiss) software.

Data processing
Data are presented as the means ± standard error of the mean (SEM) and are representative of at least three independent experiments.Statistical analyses for comparison between selected pairs were performed with Mann-Whitney test, Wilcoxon matched-pairs signed-rank test or one-way ANOVA (Tukey's multiple comparison test) as indicated using Prism v8 (GraphPad Software).

Cell culture confluency does not impact MSC multi-lineage differentiation
Previous studies showed that cell culture confluency can affect MSC attributes; however, few donors have been systematically tested in any study limiting data interpretation [8,37,38].To test the effect of cell confluency on MSC immunomodulatory functions, bone marrowderived MSCs were expanded from 6 donors (Table 1) using the same culture conditions while varying only seeding density.MSCs were then expanded for 7 days in the absence of priming cytokines prior to harvesting for co-culture or downstream analysis (Fig. 1A).Specifically, BM-MSCs were seeded at 200 cells/cm 2 and 3000 cells/cm 2 resulting in ~ 19% confluency (MSC LC ) or ~ 90% confluency (MSC HC ) at day 7, respectively, as quantified with an imaging-based algorithm (Fig. 1B and Additional file 1: Fig. S1).MSC HC proliferated slower and had a longer population doubling time (~ 4 days) compared to MSC LC (~ 2 days) in all 6 donors (Fig. 1C), consistent with contact dependent cell cycle withdrawal [39].
To determine the effect of seeding density on MSC quality, we next analysed immunophenotypic cell surface antigens endorsed by the International Society of Cell and Gene Therapy (ISCT) [40].Both MSC LC and MSC HC exhibited positive staining for CD73, CD90 and CD105, and minimal staining for hematopoietic markers CD34 and CD45 after 7 days in culture (Fig. 1D, Additional file 2: Fig. S2).The expression of CD34 and CD45 on bone marrow-derived MSCs has been  (41), which could explain their variability of these markers in our data.
To assess MSC differentiation potential in both conditions, we induced tri-lineage differentiation.Osteogenesis and adipogenesis were induced for 2 weeks as described [42,43] and cultures were stained for Alizarin Red S and Oil Red O staining, respectively.We observed positive staining of calcium-containing differentiated osteocytes and lipid droplet-containing differentiated adipocytes from all donor MSCs independent of cell seeding density.To assess chondrogenesis, the same number of MSCs from both conditions were subjected to a 3D culture pellet assay in the presence of chondrogenic differentiation media for 3 weeks [44].We observed both soluble glycosaminoglycans (sGAGs) and type II collagen in the pericellular matrix of differentiated MSCs across all donors in both conditions (Fig. 1E, F, Additional file 3: Fig. S3A, B).Together, these data indicate that culture confluency does not affect the quality or multi-lineage differentiation potential of donor MSCs.

MSC HC promotes enhanced suppression of anti-CD3/ CD28-activated T cells in co-culture
MSCs seeded at high density relative to the number of PBMCs in co-culture assays can modestly inhibit phytohaemagglutin A (PHA)-stimulated PBMC proliferation [31].These data suggest that an increased ratio of MSCs to stimulated PBMCs could impact immunomodulation during ex vivo expansion, but the target cell affected by MSC co-culture was not characterized in this study.To specifically test whether MSC confluency affects T cell proliferation, we harvested MSCs after 7 days in culture and used equal numbers of MSC LC and MSC HC for coculture with anti-CD3/CD28-activated T cells (Fig. 2A) .We first isolated CD3 + T cells from PBMCs using negative immunoselection achieving > 90% purity (Additional file 4: Fig. S4A and Table S2).
Flow cytometry showed that enriched cells comprise 58.8% and 36.6% of CD4 + T cells and CD8 + T cells, respectively (Additional file 4: Fig. S4B).We then activated the T cells with monoclonal antibodies against CD3/CD28 to mimic physiological stimulation of naïve T cells [18,33,45], in contrast to polyclonal stimulation by the plant lectin PHA [11].Activated T cells were co-cultured at the same ratio with either MSC LC or MSC HC (1:3 MSC/T cell) for 5 days across all six donors.We examined T cell proliferation after 5 days of co-culture based on established guidelines for short term lymphocyte assays [31,[46][47][48].
We first quantified the T cell proliferation index (PI) at day 5 after co-culture based on the median fluorescent intensity (MFI) from the histogram representing the CTV-stained T cells for each sample (Fig. 2B) (36).Our data show that co-culture of anti-CD3/CD28-activated T cells with MSCs led to significant suppression of proliferation in all donors as compared to activated T cells alone (positive control/PC), whereas naïve T cells showed no proliferation (negative control/NC) (Fig. 2C).Notably, co-culture with MSC HC resulted in significantly greater T cell suppression, with an average decrease of 1.90 ± 0.3165 in T cell PI compared to MSC LC in 4 of the 6 donors.Notably, our analysis shows that MSCs cultured at high confluency retain enhanced immunomodulatory capacity for at least 5 days after harvesting even in the absence of cytokine priming.

MSC HC : T cell co-cultures secrete reduced pro-inflammatory cytokines
Given the enhanced suppression of MSC HC on T cell proliferation, we next measured the levels of pro-inflammatory cytokines (IFN-ɣ and IL-17A) typically secreted by activated T cells in vivo [49,50].Media was collected from all co-culture conditions and donors at day 5 and cytokines were quantified using a bead-based Luminex assay in a replicate set of experiments.We found that all pro-inflammatory cytokines tested showed significant reduction upon co-culture with MSCs compared to activated T cells alone (positive control/PC) (Fig. 2D,  E).Notably, co-culture of MSC HC and T cells led to a significant decrease in cytokine levels compared to MSC LC for the same 4 donors that showed stronger inhibition of T cell proliferation.Our data suggest secretion of these pro-inflammatory cytokines correlates with T cell proliferation and that their levels can be more robustly modulated by MSCs cultured at high confluency.

MSC HC displays increased lactate secretion and glycolytic flux
Lactate is produced by cells that are highly glycolytic and can affect proliferation by inhibiting oxidative phosphorylation [51][52][53].Indeed, increased lactate in the media can suppress T cell proliferation and cytokine production [52,[54][55][56].Improved immunomodulatory capacity also correlates with shift towards glycolytic metabolism and lactate secretion upon priming naïve MSCs with IFN-ɣ or exposure to hypoxic conditions [56,57].To determine how MSC confluency affects lactate secretion from MSCs, we measured the lactate concentrations in the media in all conditions and donors before co-culture with T cells using a Cedex Bio Analyzer.Notably, MSCs derived from donors 1, 2, 4 and 5 that showed the highest T cell suppression activity in co-culture also showed increased lactate secretion (Fig. 3A).Using the Seahorse Flux Analyzer, we monitored the extracellular acidification rate (ECAR) measurements as an indicator of glycolysis (Fig. 3B) and showed that MSC HC displayed higher basal glycolysis compared to MSC LC in 4 of 6 donors (Fig. 3C).Although we observed a similar trend, the changes in ECAR did not correlate with changes in lactate secretion levels in some donors.It is important to note that ECAR measures the flux of protons, which can be contributed by other sources of extracellular acidification and not solely due to lactate secretion [58].Taken together, our data demonstrate that expansion of MSCs at high density prior to co-culture results in altered metabolic states that correlate with enhanced immunomodulation across multiple donors.
Because we wanted to identify potential pathways that characterize immunomodulation potency, we next categorized the donors into two groups based on MSC HC suppression of T cell proliferation (1) responders: Donor 1, Donor 2, Donor 4, Donor 5; and (2) nonresponders: Donor 3 and Donor 6.By performing DEG analysis of high confluency versus low confluency in both groups, we identified 731 DEGs that are unique to the responders (Additional file 8: Table S2), 257 DEGs that are unique to the non-responders and 1058 DEGs that are shared between the two groups (Fig. 4D).From the 731 DEGs, we performed pathway enrichment analysis and identified several biological pathways (p ≤ 0.05) associated with immunomodulation including "cytokine-cytokine receptor interaction", "cell adhesion molecules", "complement and coagulation cascade" and "PI3K-AKT signalling pathway" (Fig. 4E).Taken together, these data show that ex vivo expansion of MSCs at high cell confluency can enhance properties associated with immunosuppression suggesting our approach is a generalizable and non-invasive strategy for enriching MSCs for enhanced immunomodulatory functions (Fig. 4F).

MSC HC reduces inflammation more than MSC LC in human PBMC-induced mouse GVHD model
To examine whether MSC HC exerts greater immunosuppression than MSC LC in vivo, we investigated in a humanized mouse model of graft-versus-host disease (GVHD).Briefly, NOD-SCID IL2rg null (NSG) female mice were irradiated and injected with 5 x 10 6 human PBMCs via tail vein.We first performed a pre-dose blood withdrawal at 17 days post-PBMC injection to confirm successful immune reconstitution of human CD45 + (hCD45) cells in these GVHD-bearing mice before we administered human MSC LC , MSC HC or saline.We observed that the levels of human hCD45 cells per µl blood was in the range of 222.8 to 3215.8, with an average of 1559.4 across all PBMC-engrafted mice.This provided us an opportunity to evaluate the therapeutic effect of MSCs from the reduction in hCD45 cell count in response to MSC administration.Expanded MSCs at passage 4 were administered via intraperitoneal route into the mice 21 days post-PBMC injection.GVHD symptoms were monitored daily by observing the mice and by weekly blood withdrawal (Fig. 5A).Based on a standard clinical scoring system [59] to determine GVHD severity (Additional file 5: Fig. S5), mice administered with MSC HC had reduced body weight loss (< 10%) and experienced normal or mild decrease in movement as compared to MSC LC or saline-treated GVHD mice (Fig. 5B).The presence of hCD45 cells in blood samples indicates the extent of engraftment in the human PBMC-induced mouse GVHD model.We observed one mouse death each, in MSC LC -treated (Day 29) and saline-treated (Day 37) groups while all PBMCengrafted mice treated with MSC HC cells survived until the analysis endpoint (Day 49).We found that hCD45 cells were overall lower in MSC HC -treated GVHD mice (n = 4), as compared to saline or MSC LC treatment groups (n = 5 and n = 5, respectively) up to 14 days after MSC administration (35 days post-PBMC injection),  suggesting an initial robust effect (Fig. 5C, Additional file 6: Fig. S6).Analyses of hCD45 blood count analysis revealed that single dose of MSC HC can suppress or delay T cell proliferation by ~ 2.6-4.4-fold for 7-14 days compared to PBMC-engrafted mice treated with MSC LC or saline.After 14 days post-MSC HC treatment, we observed that the immunosuppressive effect of MSC HC began to decline as the suppression on hCD45 cell count was not observed.Nevertheless, we observed reduced lymphocyte infiltration and improved tissue integrity in MSC HC -treated GVHD mice as compared to MSC LC based on histological evaluation of the lungs, colon, liver, kidney and skin and consistent with improved survival (Fig. 5D).The onset of GVHD is triggered by human T cells from the graft that recognizes the cells of the recipient host as foreign.From histological assessment, mice treated with MSC LC and saline suffered most damage in most organs such as lungs, colon, liver, kidneys and skin, which is characterized by thickening of bronchial wall, colonic muscle wall and massive infiltration of immune cells.Thus, our results indicate that even a single dose of MSC HC can decrease inflammation and tissue damage in humanized GVHD mouse model, opening the door for further investigations of MSCs as a treatment for inflammatory diseases.

Discussion
MSCs can regulate the functions of immune cells both in vitro and in vivo; however, their clinical efficacy as an immunotherapy has been inconsistent [60].Donor heterogeneity, cell source [61] and a lack of standardized isolation and expansion protocols have posed challenges for the use of MSCs as a cell therapy.Preconditioning MSCs with pro-inflammatory cytokines or other priming approaches to improve their immunomodulatory functions have been reported [30,[62][63][64].However, priming MSCs could introduce more variability during expansion and the impact of MSC culture conditions on specific immune cell populations has not been fully investigated.Moreover, translating primed MSCs to the clinic has significant limitations due to the high cost of producing these cells as well as safety concerns related to tumorigenic potential [30].Thus, we focused on identifying conditions and mechanistic underpinnings that lead to enhanced immunomodulatory functions of basal MSCs in the context of T cell proliferation in the absence of cytokine priming.
Our study demonstrates that harvesting MSCs at high density enhanced the ability of MSCs to suppress proliferation of activated T cells in 4 out of 6 donors.Consequently, we measured levels of pro-inflammatory cytokines (IFN-ɣ, IL-17A) which were decreased in cocultures with MSC HC donors that showed the most robust effects on T cell suppression.Prior studies have demonstrated that metabolic remodelling towards enhanced glycolysis correlates with MSCs that are immunosuppressive [47,51,52], accompanied with an increased lactate production that is known to suppress immune cell function [65,66] including T cell proliferation [54,55].Notably, increased lactate production strongly correlated with the same MSC HC donors that showed enhanced immunomodulation and cytokine expression.Although lactate is considered a glycolytic by-product, it is now being recognized as a signalling molecule involved in cell survival [67,68] which could be critical for their enhanced activity in vivo.We also observed a trend for metabolic reprogramming towards glycolysis in MSC HC for most donors using the Seahorse assay.Thus, monitoring IFN-ɣ and IL-17A as well as lactate levels during MSC expansion could be used as critical quality attributes for identifying MSCs with an enhanced ability to suppress T cell proliferation.
Although several transcriptomic studies have shown that genes involved in immunomodulation are highly expressed in human MSCs cultured at high confluency as compared to low confluency [6,32,37], these studies are limited by the number of donors tested, making it difficult to identify pathways that robustly correlate with immunomodulatory activity.By performing a comprehensive transcriptome analysis across six donors, we identified several differentially regulated genes in MSC HC with known roles in immunomodulation including TRIM29 [69], BPIFB4 [70], MMP3 [71], SPP1 [72], which could hence explain enhanced immunomodulatory properties in MSC HC compared to MSC LC .By further comparing gene expression patterns in responders MSC HC Fig. 5 MSC HC reduces inflammation and disease severity in a graft-versus-host disease (GVHD) mouse model in vivo.A Timeline of treatment regimen of PBMC injection and MSCs administration.B Clinical scoring of GVHD-bearing mice with saline (n = 5), MSC LC (n = 5) and MSC HC (n = 4) with MSC administration from day 21 post-PBMC injection as indicated by green line.C Human CD45 blood count in treatment groups.D Representative histology images of lung, colon, liver, kidney and skin from each treatment group.The presence of immune cell infiltrates indicated by red arrows.Wilcoxon matched-pairs signed-rank test was performed between selected pairs.*p ≤ 0.05; *** p ≤ 0.001; ns represents non-significant.There were originally 15 mice for the study, but human CD45 engraftment was not successful in one mouse on Day 17 post-PBMC injection, hence to be removed from the study (See figure on previous page.)(Donor 1, 2, 4 and 5) which exhibit enhanced immunosuppression towards T cell proliferation compared to non-responders (Donor 3 and 6), we observed an enrichment of pathways regulating immune response including cytokine-cytokine receptor interaction pathway, cell adhesion pathway and PI3K-AKT signalling pathway.Within the cytokine-cytokine receptor interaction pathway, genes associated with MSC-mediated immunomodulation such as BMP4 [73], CCL26 [74], CX3CL1 [75,76], IL6 [77], LEP [78] and TGFBR1 [79] were identified.We also observed the differential expression of SDC1 [80], which was highlighted in the cell adhesion pathway that mediates interaction between MSCs and immune cells.It is known that the activation of PI3K-AKT signalling pathway confers anti-inflammatory functions to MSCs [81], which suggests that its enrichment in the responders may explain the enhanced immunosuppression exhibited by MSC HC .Overall, the ability to control expression of these genes to enhance MSC-mediated immunomodulation functions provides an exciting area to develop approaches to overcome the challenges of MSC therapy, which are donor heterogeneity and variations to MSC properties during ex vivo expansion.
Steroid-based anti-inflammatory therapy is often used as first line therapy, but developing resistance to steroids is frequently observed in patients and lead to high mortality [14,19,23].The therapeutic potential of MSCs offers an alternative treatment for steroid-refractory patients.The ability to define culture conditions that enrich for MSCs with enhanced immunomodulatory activity could facilitate their use as a treatment of diseases characterized by pro-inflammatory T cells like GVHD [82].Collectively, our in vivo data support our in vitro findings demonstrating that MSC HC can suppress T cell proliferation in peripheral blood.Administration of even a single dose at Day 21 significantly reduced the GVHD clinical score, prolonged survival, reduced immune cell infiltration to organs and organs damage compared to MSC LC and saline (non-MSC)-treated mice.Thus, results from the peripheral blood count, GVHD clinical score and histopathological analyses support that MSC HC suppresses T cell proliferation and inflammation in a humanized murine models of xenogeneic GVHD.
Although the immunosuppressive effect of MSC HC cells began to wane after 14 days post-MSC HC treatment, we observed a significant decrease in lymphocyte infiltration in the organs and improved survival.Thus, it is possible that multiple dosing of MSC HC cells to PBMCengrafted mice every 14 days or increasing the number of MSC HC injected could prolong the improvement of GVHD clinical score.

Conclusion
The potential of MSCs in cell therapy is often limited by donor heterogeneity and differences in cell culture protocols across laboratories.Taken together, we show that enrichment of immunosuppressive functions during ex vivo expansion of MSCs can be achieved by culturing at high confluency.Future studies of MSC HC as a treatment of steroid-resistant GVHD or whether a maximal effect can be achieved in combination with existing drugs such as prednisone and ibrutinib [83] warrants further investigation that could greatly impact the lives of many patients.

Fig. 1
Fig. 1 Phenotypic characterization of bone marrow-derived MSCs in this study.A Method for varying low (MSC LC ) and high confluency (MSC HC ) of BM-MSCs from 6 donors.B Representative images of MSC LC and MSC HC for Donor 1 (scale bar: 1 mm).C Population doubling time of MSC LC and MSC HC for Donor 1 to 6. D Flow cytometry-mediated characterization of surface markers to determine % positive BM-MSCs in each experimental condition for this study.E In vitro differentiation of MSC LC and MSC HC from Donor 1. Alizarin red staining for calcium deposits detected in osteoblasts, Oil red O staining for oil droplets in adipocytes and collagen II staining was performed for chondrocytes.Scale bar: 100 μm.F Differentiation assays performed in technical triplicates and quantified for Donor 1 as compared to undifferentiated cells (UD).One-way ANOVA (Tukey's multiple comparison test) or Mann-Whitney test was performed between selected pairs as statistical test.*p ≤ 0.05; **p ≤ 0.01 *** p ≤ 0.001; ns represents non-significant

(Fig. 2
See figure on next page.)Inhibition of T cells proliferation in the presence of co-culture with MSC LC or MSC HC .A CD3 + T cells were isolated from human PBMCs and stained with Celltrace Violet (CTV) proliferation dye to track T cell proliferation.Co-culture of T cells with MSC LC or MSC HC from Donor 1 to 6 was performed with CTV-stained T cells and anti-CD3/CD28 antibodies (Abs).B Representative histogram plots of CD3 + T cell proliferation in co-culture with MSCs, assessed by CTV dilution.C The extent of T cell proliferation was determined with T cell proliferative index (PI) calculated from their respective histogram by the following formula: PI = Log[FI nd /MFI all ]/Log[2], with MFI all = median fluorescence intensity of all viable T cells and FI nd = peak fluorescence intensity of the viable non-divided cells.White bar indicates co-culture of MSCs harvested at low confluency (MSC LC ) with T cells; light grey bar indicates co-culture of MSCs harvested at high confluency (MSC HC ) with T cells; black bar indicates positive control (PC) containing anti-CD3/CD28-activated human CD3 + T cells without MSCs; dark grey bar indicates negative control (NC) containing naïve CD3 + T cells.The concentration of D IFN-ɣ and E IL-17A secreted into the co-culture media were measured by ELISA.Data shown were expressed as mean ± SEM of triplicates of separate experiments, representative of 6 donors.One-way ANOVA (Sidak's multiple comparison test) was performed between selected pairs in each donor.*p ≤ 0.05; ** p ≤ 0.01 *** p ≤ 0.001; ns represents non-significant

Fig. 3 Fig. 4
Fig. 3 MSC HC produced more lactate than MSC LC .A Lactate production rate were obtained from Cedex measurements and displayed for 6 donors.B Cell number-normalized ECAR profile plots of MSC LC and MSC HC from Donor 1 in response to glucose, oligomycin and 2-deoxyglucose (2-DG) in glycolysis stress test.C Basal glycolysis was calculated as (maximum rate measurement before Oligomycin injection)-(last rate before Glucose injection) for each donor.Data shown were expressed as mean ± S.E.M. of three measurements, from two independent experiments.Unpaired t-test (two-tailed) was performed.*p ≤ 0.05; **p ≤ 0.01; ns represents non-significant

Table 1
Description of MSC donorsBone marrow-derived MSCs from human donors.Y years, M male, F female